Mirai’s Inventory Management System

Discipline: CS

Advisor: Richard Leinecker

Mirai Arcade, a Japanese claw-cade in Orlando, manages 1,200+ products across 8 storage locations with 20+ employees making hundreds of daily inventory updates on a legacy system plagued by poor data modeling, no unique product IDs, slow full-table scans, and intuition-based restocking. Our team rebuilt this from the ground up as a modern, event-driven inventory platform. The system features a normalized PostgreSQL database via Supabase with standardized product identification, location-specific inventory tables, and a complete audit trail for accountability. A Java/Spring Boot backend handles secure REST operations using the Transactional Outbox pattern to guarantee data consistency, while Apache Kafka streams inventory events to two Python microservices: a Forecasting Service that calculates real-time stockout timelines and optimal reorder quantities using a demand-variance algorithm, and a Messaging Service that pushes threshold alerts directly to Slack. The React/TypeScript frontend delivers live inventory updates without requiring a page refresh. The entire stack runs in Docker containers on DigitalOcean, making it scalable, maintainable, and production-ready. Unlike generic inventory tools, this system is purpose-built for Mirai's exact operational workflow, replacing intuition-based decision making with data-driven reorder recommendations that prevent stockouts, improve employee accountability, and establish a reliable foundation for long-term business growth.

Back to Showcase